Scribd
Upload
27 views4 pages

How To Conduct Today Amamoto

This document provides instructions for conducting a Toda-Yamamoto Granger causality analysis in two steps. First, it explains how to derive the vector autoregressive (VAR) equations needed for the analysis based on the maximum order of integration and optimal lag length. Second, it outlines the step-by-step process for running the analysis using an ordinary least squares approach and interpreting the results to determine causality direction between variables.

Uploaded by

Alok Tiwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
27 views4 pages

How To Conduct Today Amamoto

This document provides instructions for conducting a Toda-Yamamoto Granger causality analysis in two steps. First, it explains how to derive the vector autoregressive (VAR) equations needed for the analysis based on the maximum order of integration and optimal lag length. Second, it outlines the step-by-step process for running the analysis using an ordinary least squares approach and interpreting the results to determine causality direction between variables.

Uploaded by

Alok Tiwari
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/345948326

HOW TO CONDUCT/RUN A TODA YAMAMOTO GRANGER CAUSALITY ANALYSIS

Method · November 2020


DOI: 10.13140/RG.2.2.29928.90881

CITATION READS
1 3,987

1 author:

Stephen Ekeke
Pukyong National University
3 PUBLICATIONS   2 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

THE IMPACT OF THE NON-OIL EXPORT SECTOR OF NIGERIA AND ITS IMPACT ON ECONOMIC GROWTH View project

All content following this page was uploaded by Stephen Ekeke on 16 November 2020.

The user has requested enhancement of the downloaded file.


HOW TO CONDUCT/RUN A TODA YAMAMOTO GRANGER CAUSALITY
ANALYSIS.
By

EKEKE STEPHEN

This document would try to explain how to derive your Toda-Yamamoto granger causality
equations and also how to conduct the analysis.
1. How to derive equations
2. How to run the analysis

1. How to derive equations

The ARDL co-integration bounds test that shows the relationship of the variables on the
short and long run, does not indicate the direction of the causation among the variables. This makes
the estimation results incomplete. Granger (1969) stated that the aim for estimation results to
undergo the causality technique was not to find the relationship between the variables but to test
the causality between them. This strengthens the results of the estimation because it verifies the
causation effect among the variables.

For example, if Variable A (lnVa) is positively related to Variable B (lnVb), the Granger
causality test is necessary to figure out what causality direction exist between them. What the
causality test does is to determine if the causality runs from Va to Vb (Va →Vb); Vb to Va
(Vb→Va); if Va and Vb both caused each other (Va↔Vb); and if the causality between Va and
Vb does not exist. To execute this Granger causality test, the version of Toda-Yamamoto is more
reliable because it is justifiable regardless of if the variables are not co-integrated or co-integrated
at a random order or with the orders of I(0), I(1), respectively.

The Toda-Yamamoto Granger Causality approach adopted a revised Wald test for
restriction on each parameters of the Vector Auto Regression VAR (k), where the lag length is k.
The actual order of the system (k) is supplemented by the highest order of integration (dmax). The
VAR (k + dmax) is estimated with the coefficients of the last lagged dmax vector being ignored. The
Wald statistic adopts a chi-square allocation of a function with degrees of freedom that corresponds
to the number of eliminated lagged variables.

The empirical model in the Vector Autoregressive (VAR) system to execute Toda-
Yamamoto’s approach to Granger causality test is as follows for example: If you have X number
of variables namely “Variables a-g” with Variable B (Vb) as the dependent variable, the VAR
representation is listed below.
𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

𝑉𝑏𝑡 = 𝛼0 + ∑ 𝛼1𝑖 𝑉𝑏𝑡−𝑖 + ∑ 𝛼2𝑖 𝑉𝑏𝑡−𝑗 + ∑ 𝛽1𝑖 𝑉𝑎𝑡−𝑖 + ∑ 𝛽2𝑖 𝑉𝑎𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝛾1𝑖 𝑉𝑐𝑡−𝑖 + ∑ 𝛾2𝑖 𝑉𝑐𝑡−𝑗 + ∑ 𝛿1𝑖 𝑉𝑑𝑡−𝑖 + ∑ 𝛿2𝑖 𝑉𝑑𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝜌1𝑖 𝑉𝑒𝑡−𝑖 + ∑ 𝜌2𝑖 𝑉𝑒𝑡−𝑗 + ∑ 𝜎1𝑖 𝑉𝑓𝑡−𝑖 + ∑ 𝜎2𝑖 𝑉𝑓𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝜇1𝑖 𝑉𝑔𝑡−𝑖 + ∑ 𝜇2𝑖 𝑉𝑔𝑡−𝑗 𝑢1𝑡 ……Equation 1


𝑖=1 𝑗=𝑘+1

Please note however….., you must repeat the VAR equation for the same amount of
variables that you have in your ARDL equation. Like below (please take note how the variable
representation change in equation 2. Repeat this for each other variables):

𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

𝑉𝑎𝑡 = 𝛼0 + ∑ 𝛽1𝑖 𝑉𝑎𝑡−𝑖 + ∑ 𝛽2𝑖 𝑉𝑎𝑡−𝑗 + ∑ 𝛼1𝑖 𝑉𝑏𝑡−𝑖 + ∑ 𝛼2𝑖 𝑉𝑏𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝛾1𝑖 𝑉𝑐𝑡−𝑖 + ∑ 𝛾2𝑖 𝑉𝑐𝑡−𝑗 + ∑ 𝛿1𝑖 𝑉𝑑𝑡−𝑖 + ∑ 𝛿2𝑖 𝑉𝑑𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥 𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝜌1𝑖 𝑉𝑒𝑡−𝑖 + ∑ 𝜌2𝑖 𝑉𝑒𝑡−𝑗 + ∑ 𝜎1𝑖 𝑉𝑓𝑡−𝑖 + ∑ 𝜎2𝑖 𝑉𝑓𝑡−𝑗


𝑖=1 𝑗=𝑘+1 𝑖=1 𝑗=𝑘+1

𝑘 𝑑𝑚𝑎𝑥

+ ∑ 𝜇1𝑖 𝑉𝑔𝑡−𝑖 + ∑ 𝜇2𝑖 𝑉𝑔𝑡−𝑗 𝑢1𝑡 ……Equation 2


𝑖=1 𝑗=𝑘+1

2. How to run the analysis


Performing the co-integration tests and the diagnostic test is not sufficient to describe the
direction of the causality flow among the variables. For the Toda-Yamamoto test for Granger
causality, you have to determine the maximum order of integration (dmax) by use of the ADF unit
root test. Once this is obtained, you generate a VAR representation of equations to obtain the
optimal lag (k) of the variables.
By adopting the procedure of Lutkepohl (2005), you then add the maximum order of
integration and the optimal lag to determine the lag length of the variables. By means of ordinary
least squares, you connect this lag length number to the number of endogenous variables in the
VAR sample size represented in your equations. This thereby analyses the variables as exogenous
variables.
Below is a step-by-step procedure to run the Toda-Yamamoto Granger Causality analysis.
1. Determine the maximum order of integration when you conduct the ADF unit root test (for
example Order 1)
2. Select all your variables accordingly and open together as VAR.
3. A dialogue box opens “VAR SPECIFICATION” – just click OK.
4. A new dialogue box opens. Click on “VIEW”, the “LAG STRUCTURE”, then “LAG
LENGTH CRITERIA”.
5. A new dialogue box opens “LAG SPECIFICATION”. Here you specify how much “LAGS
TO INCLUDE”. Type in a number and once you get the best number, you click OK.
6. A new box opens “VAR LAG ORDER SELECTION CRITERIA”. This box has various
criteria (LAG, LOGL, LR, FPE, AIC, SC, HQ). Each criteria has various numbers up to the
number of lags you included. Look for the row in which the lag number has an (*) sign. Most
notably look at AIC, SC, and HQ criteria. The row number that has the asterisk (*) sign
becomes your optimal lag to use in the VAR equation box. For example, this number is (2).
7. Go over to the menu bar and click on “ESTIMATE”. Please note do not forget the optimal lag
number.
8. After clicking on “ESTIMATE”, a “VAR SPECIFICATION” box opens again.
9. At the part title “LAG INTERVALS FOR ENDOGENOUS”, enter your optimum lag number
which for example maybe 2 (refer to step 6 to confirm this number). To add this number, delete
the last number in the box and replace it with your new number in this case could be 2, such
as it becomes [1 2].
10. After doing this, below has a box titled “ENDOGENOUS VARIABLES”. Copy the equation
specification that is specified above the dialogue box in “ENDOGENOUS VARIABLES”.
Copy this equation to the “EXOGENOUS SECTION” and paste it after the constant (c).
11. Beside each variable in the EXOGENOUS SECTION, type an open and close bracket [c]. In
each bracket type in the minus sign and the summation number of your optimal lag number
(which in this could be 2) and your maximum lag number (which in this case can be 1). Such
that each variable has (-3) after it. Then click OK.
12. In the results page, click on “VIEWS”. Then click on “LAG STRUCTURE” and then click on
“GRANGER CAUSALITY/BLOCK EXOGENEITY TESTS”.
13. From the next box that opens, you can now determine the causal direction of each variable by
identifying to ensure the probability statistics is significant at either (1%, 5%, 10%). If two
variables are both significant, then you have a “BI-DIRECTIONAL CAUSALITY” meaning
that Variable A causes Variable B and vice versa. If only one variable is significant, then you
have a “UNI-DIRECTIONAL CAUSALITY” meaning that Variable A causes Variable B, but
Variable B did not cause Variable A. If both variables are not significant then there was no
causal relationship between them.
I hope this document can help you successfully run your Toda-Yamamoto Granger Causality
Analysis.
Cheers!!!!
Ekeke Stephen

View publication stats

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy